Chatbots may be 'racist' in health-related queries

Researchers from Stanford University asked medical questions to chatbots including OpenAI's "ChatGPT" and "GPT-4," Google's "Bard," and the AI firm Anthropic's "Claude" in order to detect race-based differences in their responses.

EVEN POPULAR PLATFORMS LIKE TWITTER AND SIMILAR ONES HAVE THIS RACISM

In addition to asking the chatbots questions about topics such as "lung capacity, skin thickness, and pain threshold of Black men and women," the same questions were asked again without specifying race to identify differences in the answers.

The robots' responses contained "long-held medical myths" regarding race-based biological differences.

The researchers stated that the false beliefs that historically influenced doctors' diagnoses of Black patients' conditions were also present in the robots' responses.

The study was published in the journal "Nature Digital Medicine."